Disproof of the Odd Hadwiger Conjecture
Marcus Kühn, Lisa Sauermann, Raphael Steiner, Yuval Wigderson
TL;DR
This work disproves the odd Hadwiger conjecture by constructing, via a careful two-layer overlay of triangle-free random graphs and a complement operation, an n-vertex graph G with α(G) ≤ 2 and χ(G) ≥ (3/2 − δ)n that does not contain K_t as an odd minor for large t. The authors develop a probabilistic framework with balanced random graphs and ε-respectful pairings, and prove that no large odd connected pairing can occur w.h.p., which blocks large odd minors. Central to the argument are two-layer blowup constructions, precise control of triangles, and a cascade of concentration and coupling lemmas that together bound the existence of odd minors. The results not only refute the conjecture in a strong form but also clarify the tightness of the approach, illustrating a fundamental barrier at the 3/2 factor for graphs with independence number 2.
Abstract
We prove that there exist graphs which do not contain $K_t$ as an odd minor and whose chromatic number is at least $(\frac 32-o(1))t$. This disproves, in a strong form, the odd Hadwiger conjecture of Gerards and Seymour from 1993.
